比較 VGG, resnet和inception的圖像分類效果
阿新 • • 發佈：2019-02-24
eat dap pri 比較 分類 pad 兩層 init 效果
簡介
VGG, resnet和inception是3種典型的卷積神經網絡結構。
- VGG采用了3*3的卷積核，逐步擴大通道數量
- resnet中，每兩層卷積增加一個旁路
- inception實現了卷積核的並聯，然後把各自通道拼接到一起
簡單起見，直接使用了[1]的代碼來測試 resnet，然後用[2],[4]的代碼替換[1]中的model，改了改通道，測 VGG 和 inception。
GPU是gtx1050，主板開始是 x79，後來壞了，換成 x470，GPU占比提高很多。
CPU占比始終100%
實驗結果
超參數：epochs=80,lr=0.001,optim=Adam
數據集：cifar10
| _ | 參數個數(k) | 訓練時間(m) | 精度(%) | GPU內存(M) | GPU占比(%) |
|---|---|---|---|---|---|
| resnet | 195 | 22 | 88 | 607 | 99 |
| vgg_bn | 207 | 17 | 84 | 535 | 60 |
| inception | 107 | 19 | 80 | 613 | 98 |
結論：條條道路通羅馬。
附加實驗
因為方便，註釋掉 Batch Normalization，以及 Data Augmentation 又試了兩次。
| _ | 參數個數(k) | 訓練時間(m) | 精度(%) | GPU內存(M) | GPU占比(%) |
|---|---|---|---|---|---|
| resnet | 195 | 22 | 88 | 607 | 99 |
| resnet-BN | 195 | 19 | 86 | 553 | 99 |
| resnet-DA | 195 | 22 | 64 | 607 | 99 |
結論：Data Augmentation很重要
代碼改動
class ResNet(nn.Module): def __init__(self, block, layers, num_classes=10): super(ResNet, self).__init__() self.in_channels = 16 self.conv = conv3x3(3, 16) self.bn = nn.BatchNorm2d(16) self.relu = nn.ReLU(inplace=True) self.layer1 = self.make_layer(block, 16, layers[0]) self.layer2 = self.make_layer(block, 32, layers[1], 2) self.layer3 = self.make_layer(block, 64, layers[2], 2) self.avg_pool = nn.AvgPool2d(8) self.fc = nn.Linear(64, num_classes) print('# generator parameters:', sum(param.numel() for param in model.parameters()))
class VGG(nn.Module):
def __init__(self, features, num_classes=10, init_weights=True):
super(VGG, self).__init__()
self.features = features
self.avgpool = nn.AdaptiveAvgPool2d((3, 3))
self.classifier = nn.Sequential(
nn.Linear(9 * 8 * 8, 64),
nn.ReLU(True),
#nn.Dropout(),
nn.Linear(64, 64),
nn.ReLU(True),
#nn.Dropout(),
nn.Linear(64, num_classes),
)
def vgg_bn(**kwargs):
cfg = [16, 16, 'M', 32, 32, 'M', 32, 32, 'M', 64, 64, 'M', 64, 64, 'M']
model = VGG(make_layers(cfg, batch_norm=True), **kwargs)
class Inception_v1(nn.Module):
def __init__(self, num_classes=10):
super(Inception_v1, self).__init__()
#conv2d0
self.conv1 = conv3x3(3, 6)
self.max_pool1 = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.lrn1 = nn.BatchNorm2d(6)
self.inception_3a = Inception_base(1, 6, [[16], [16,32], [8, 16], [3, 16]]) #3a
self.inception_3b = Inception_base(1, 80, [[40], [32,48], [12, 16], [3, 16]]) #3b
self.max_pool_inc3= nn.MaxPool2d(kernel_size=3, stride=2, padding=0)
self.inception_5a = Inception_base(1, 120, [[40], [32,48], [12, 16], [3, 16]]) #5a
self.inception_5b = Inception_base(1, 120, [[40], [32,48], [12, 16], [3, 16]]) #5b
self.avg_pool5 = nn.AvgPool2d(kernel_size=3, stride=2, padding=0)
self.dropout_layer = nn.Dropout(0.4)
self.fc = nn.Linear(120*9, num_classes)
引用
[1] https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/02-intermediate/deep_residual_network/main.py
[2] https://github.com/pytorch/vision/blob/master/torchvision/models/vgg.py
[3] https://github.com/pytorch/vision/blob/master/torchvision/models/resnet.py
[4] https://github.com/antspy/inception_v1.pytorch/blob/master/inception_v1.py
比較 VGG, resnet和inception的圖像分類效果